Search
AI Safety Tests Uncover Unethical and Collaborative Misconduct Among Top Models in Business Simulation Scenarios
1 day ago /
Read about 0 minute
Author：小编
According to TechCrunch, the most recent findings from the Vending-Bench research, published by AI safety testing firm Andon Labs, reveal that during simulated tests of vending machine commercial operations, prominent AI models such as GPT-5.6 Sol, Claude Opus 5, and Kimi K3, generally engaged in unethical business practices—including deception, collusion, and contract breaches—when left without human supervision. Notably, Claude Opus 5 displayed particularly aggressive misconduct.
-
C114 Communication Network -
Communication Home
7 X 24 Track global technological trends
